This comment addresses two inferences used by Johnson (2009) to argue for porphyroblast rotation during bulk coaxial shortening. Firstly, he interprets that porphyroblast inclusion trails that are inclined (his figure 1) became inclined because the porphyroblast rotated during growth. He used orientation data from millipede microstructures from one hand sample and concluded on the basis of the total spread of inclusion trail orientations and porphyroblast axial ratios that porphyroblasts had rotated relative to one another during ductile deformation. Secondly, he presents a numerical model that indicates that asymmetrically shaped porphyroblasts rotated during coaxial deformation. However, I will show that the porphyroblasts did not rotate after they grew, in spite of localized shearing along the developing S3 that rotated visible matrix S2 in Johnson's figure 1. Rather, they grew over a foliation that was inclined before they nucleated. Furthermore, I will show that numerical modeling can produce both porphyroblast rotation and porphyroblast non-rotation just by modifying the rheology.

Figure 1 in Johnson (2009) features two plagioclase porphyroblasts from the same sample as that described in detail by Bell and Bruce (2007; the samples used in both papers were collected as adjacent pieces by Tim Bell [2009, personal commun.]). Bell and Bruce (2007) show a porphyroblast with an S2 foliation (their figures 9 and 10) that is much more rotated than that featured by Johnson in his figure 1. Bell and Bruce's excellent description of microstructural changes of S2, from inside the porphyroblast to the matrix in its strain shadow in Q-domains, reveals that porphyroblast growth post-dated rotation of the S2 foliation. This is certain because the foliation preserved inside the porphyroblast and in its strain shadow was reactivated during rotation. Sufficient clockwise shearing occurred on S2 during reactivation prior to porphyroblast growth, to reverse the original anticlockwise asymmetry of S1, crenulated against S2 preserved in the un-reactivated matrix of the adjacent fold hinge (see figure 10 in Bell and Bruce, 2007). Clearly, differential rotation of S2 within porphyroblasts from this sample is not an indication that the porphyroblasts have rotated.

I used the commercial software Abaqus (Dassault Systémes Simulia Corp.) that uses finite element analysis to simulate porphyroblast rotation/non-rotation during bulk coaxial deformation, which can of course lead to localized non-coaxial deformation (e.g., Fay et al., 2008). This analysis assumed a Drucker-Prager rheology with velocity boundary conditions at the top and bottom of the model (Fig. 1A) until 50% shortening was produced at a strain rate of 10−12/s. The experiment used porphyroblasts that are shaped similar (Fig. 1A) to those described in Johnson (2009) so that the results could be directly contrasted. For a competency contrast of 10 with the matrix, no rotation of the porphyroblasts occurred (Fig. 1B). The competency contrast between the matrix and porphyroblasts was progressively reduced until 1.5° of clockwise rotation of the unsymmetrically shaped porphyroblast was produced (Fig. 1C) at a competency contrast of 2. Similar results were obtained using a Mohr-Coulomb or a viscoelastic material definition or with combined coaxial and non-coaxial deformation. These results emphasize the vital role of the modeling approach and the matrix-porphyroblast competency contrast in producing non-rotation of porphyroblasts (e.g., Bell and Bruce, 2007). Fay et al. (2008, 2009) showed that both porphyroblast rotation and non-rotation are possible during combined coaxial and non-coaxial deformation, unless millipede microstructures are produced, when any rotation ceases. Clearly, the microstructures and numerical modeling presented by Johnson (2009) can be as easily interpreted and modeled to show non-rotation. Therefore, the debate is far from being settled.